Seyedpouyan Ahmadpanah
A Thesis
In
The Department
Of
Building, Civil and Environmental Engineering
Presented in Partial Fulfilment of the Requirements
For the Degree of Master of Applied Science in Civil Engineering at
Concordia University
March 2017
Abstract
Numerical Study of Two-Phase Turbulent Flow in Hydraulic Jumps
Seyedpouyan Ahmadpanah
Hydraulic jump is a rapidly varied flow phenomenon that the flow changes suddenly from supercritical to subcritical. Hydraulic jumps are frequently observed to exist in natural river channels, streams, coastal water, and man-made water conveyance systems. Because of a sudden transition of flow regime, hydraulic jumps result in complex flow structures, strong turbulence, and air entrainment. Accordingly, they are two-phase flow, with air being the gas phase and water being the liquid phase. Consequences of the occurrence of hydraulic jumps include: unwanted fluctuations in the water surface with unstable waves and rollers, undesirable erosion of channel sidewalls and channel bottom, and reduced efficiency for water conveyance systems. Thus, it is important to study various aspects of the phenomenon.
So far, knowledge of the phenomenon is incomplete. The main objective of this research is to improve our understanding of the complex flow structures and distributions of air entrainment in a hydraulic jump. Previously, both experimental and computational studies of the phenomenon have typically suffered a scale problem. The dimensions of the setup being used were unrealistically too small.
In this research, we took the computational fluid dynamics (CFD) approach, and simulated hydraulic jumps at relatively large and practical dimensions. This would help reduce artificial scale effects on the results. On the basis of Reynolds averaged continuity and momentum equations, CFD simulations of hydraulic jumps were performed for four different cases in terms of the approach flow Froude number Fr1, ranging from 3.1 to 5.1. The Reynolds number is high (between 577662 and 950347), which ensures turbulent flow conditions. The CFD model channel is discretized into 2,131,200 cells. The mesh has nearly uniform structures, with fine spatial resolutions of 2.5 mm. The volume of fluid method provides tracking of the free surface. The standard k-ε turbulence model provides turbulence closure.
For each of the simulation cases, we carried out analyses of time-averaged air volume fraction, time-averaged velocity, time- and depth-averaged (or double averaged) air volume fraction at a series of locations along the length of the model channel (Note that the terms air volume fraction
and void fraction are used interchangeably in this thesis). We compared the CFD predictions of air volume fraction with available laboratory measurements. It is important to note that these measurements were made from laboratory experiments that corresponded to essentially the same values of Fr as this CFD study, but used a channel of smaller dimensions, in comparison to the CFD model channel. The CFD results of time-averaged air volume fraction are reasonable, when compared to the experimental data, except for the simulation case with Fr1 = 3.8. For all the four simulation cases, the predicted variations in air volume fraction show a trend in consistency with the experimental results. For the three simulation cases (with Fr1 = 3.1, 3.8 and 4.4), the time-averaged air volume fraction in the hydraulic jumps is larger at higher Reynolds number. However, for the simulation case with Fr1 = 5.1, it is smaller at higher Reynolds number. This implies that the amount of air being entrained into a hydraulic jump depends on not only Fr1 but also the depth of the approach flow. In future studies of the hydraulic jump phenomenon, one should consider using approach flow of realistically large dimensions at various values of Fr1, for realistic predictions of air entrainment in hydraulic jump rollers.
Acknowledgment
I would like to express my gratitude to my supervisor, Dr Samuel Li. His guidance, encouragement and support throughout the research are greatly appreciated. His abundant help and invaluable assistance, support and guidance in preparing my thesis, are especially appreciated. I am truly fortunate to have such an excellent supervisor.
I would like to thank my dear parents and sister for their endless love and support through the whole duration of my studies.
Table of Contents
List of Figures... ix
List of Tables ... xiv
List of Symbols... xv
Chapter 1 Introduction ... 1
1.1 Background ... 1
1.2 Specific aims of this research work... 6
1.3 Scope of this research work ... 7
1.4 Highlights of research contributions ... 7
Chapter 2 Review of the Pertinent Literature ... 9
2.1 The classical theory of hydraulic jumps ... 9
2.2 Hydraulic jump in rectangular channels... 10
2.3 Basic characteristics of hydraulic jumps in rectangular channels ... 11
2.4 Similitude of models for hydraulic jumps ... 13
2.4.1 Viscous force – Reynolds number law ... 13
2.4.2 Gravity force – Froude number law ... 14
2.4.3 Surface tension – Weber number law ... 14
2.5 Experimental studies of air entrainment in hydraulic jump ... 14
2.6 Numerical studies of air entrainment in hydraulic jumps ... 22
2.7 Summary ... 28
Chapter 3 Modelling Methodology ... 29
3.1 Model domain and geometry... 29
3.2 Volume of fluid (VOF) model theory ... 30
3.2.1 Steady – state and transient VOF calculations ... 30
3.2.3 Material properties ... 32
3.3 Reynolds averaged Continuity and Momentum equation ... 32
3.4 Energy equation... 33
3.5 Surface tension ... 33
3.6 Standard 𝒌-ε turbulent model ... 34
3.6.1 Transport equations for the standard 𝒌-ε model ... 34
3.7 Finite volume mesh ... 35
3.8 Boundary conditions ... 37 3.8.1 Pressure inlet... 37 3.8.2 Wall ... 38 3.8.3 Velocity inlet ... 39 3.8.4 Pressure outlet... 40 3.9 Initial condition ... 41
Chapter 4 Results and Discussions... 44
4.1 Introduction ... 44
4.2 Air-water flow structure in hydraulic jumps ... 44
4.2.1 Inflow condition ... 44
4.2.2 Air entrainment and flow pattern ... 45
4.2.3 Air volume fraction profile ... 47
4.3 CFD simulation cases ... 48
4.4 Predicted flow field and air volume distribution ... 49
4.4.1 Simulation case 1 ... 50
4.4.2 Simulation case 2 ... 60
4.4.3 Simulation case 3 ... 70
4.5 Summery ... 87
Chapter 5 Conclusions ... 90
5.1 Concluding remarks... 90
5.2 Suggestions for future research ... 91
List of Figures
Figure 1.1 Photo of the experimental setup in Concordia University’s Water Resources
Engineering Laboratory: (a) a rectangular recirculation flume; (b) hydraulic jump as bubbly flow. The direction of flow was from right to left. A control sluice gate allowed a small opening and produced supercritical flow downstream of itself. At the downstream end of the flume, a control tail gate was raised, producing subcritical flow upstream of itself. The two controls created the hydraulic jump. There is an exchange of air mass across the flow surface, with a net entrainment of air mass into the flow, forming air bubbles in the flow. ... 2 Figure 1.2 Burdekin dam on the Burdekin River in Queensland, Australia, showing hydraulic jumps induced by obstructions and a gradient change. (http://www.abc.net.au/news/2016-05-26/the-burdekin-falls-dam-spills-over-as-cyclone/7446962, accessed on February 26, 2017) ... 3 Figure1.3 Hydraulic jump on the Naramatagawa River's stream. (https://commons.wikimedia.org/wiki/File:Hydraulic_jump_on_Naramatagawa_Ri ver's_stream.JPG, accessed on February 27, 2017) ... 3 Figure1.4 Hydraulic jump downstream of a weir in an open channel. (http://www.itrc.org/projects/flowmeas.htm, access on september 2, 2016) ... 4 Figure 2.1 Specific energy diagram, hydraulic jump and specific force diagram (adopted from Houghtalen et al. 2017). ... 9 Figure 2.2 Classifications of hydraulic jumps according to the upstream Froude number Fr1
(adopt from US Bureau of Reclamation 1987). ... 12 Figure 2.3 Definition diagram of hydraulic jump experiments in a rectangular channel (adopted from Chanson and Brattberg, 2000). Note that V represents the flow velocity in the
x-direction. ... 15 Figure 2.4 Vertical distributions of air concentration (air volume fraction) and bubble frequency in hydraulic jump rollers (from Chanson and Brattberg 2000). ... 17 Figure 2.5 Distributions of air volume fraction and bubble count rate, measured at 0.3 m downstream of the jump toe by Chanson (2006). The upstream Froude number was Fr1 = 8.6. The channel was 0.5 m wide (adopted from Chanson 2006). ... 18
Figure 2.6 Vertical profile of air volume fraction at (x - x1)/y1 = 11.63. The experimental conditions were Fr1 = 8.37, Re1 = 38410, and y1 = 0.0129 m (adopted from Gualtieri and Chanson 2007) ... 19 Figure 2.7 Comparison of air volume fraction in hydraulic jumps between Chanson (2006) (with Re1 = 24738 and 68900) and Chanson and Murzyn (2008) (with Re1 = 38576). The upstream Froude number is Fr1 = 5.1. The distance is (x - x1)/y1 = 8 (adopted from Chanson and Murzyn 2008). ... 20 Figure 2.8 Vertical distributions of air volume fraction at a series of positions along the length
of the hydraulic jump. The flow conditions were: Q = 0.0347 m3/s, y
1 = 0.0206 m, x1 = 0.8 3m, Fr1 = 7.5, and Re1 = 68000 (adopted from Wang and Chanson 2015).... 22 Figure 2.9 Distributions of air volume fraction predicted with the Reynolds-averaged equation model (left) and Detached Eddy Simulation model (right) at Fr1 = 1.98 at (xx1)/y1 = 0.85, 1.7 and 2.54 (from top to bottom), in comparison with the measurements of Murzyn et al. (2005). The middle column presents DES results accounting for contributions from bubbles, while excluding those from the wavy interface (adopted from Ma et al. 2011). ... 23 Figure 2.10 Model domain, mesh and boundary conditions in Xiang et al. (2014). ... 24 Figure 2.11 Predicted water velocity vectors for Case 3 (Table 2.4) in Xiang et al. (2014). ... 25 Figure 2.12 Contours of water and air volume fraction for: (a) Case 1, and (b) Case 2 in Xiang et al. (2014). ... 25 Figure 2.13 Distributions of air volume fraction for Case 1 at axial sections (xx1)/y1= 0.91 m (panel a), 1.7 (panel b), and 3.41 (panel c). Note that y is the vertical position (adopted from Xiang et al. 2014). ... 26 Figure 2.14 Distributions of predicted volume fraction for a 2-D simulation with the upstream Froude number Fr1 = 4.82. Panel (a) shows an instantaneous distribution. Panels (b), (c), (d), (e), and (f) show the time-averaged distributions over the durations of 1, 5, 10, 15, and 20 s, respectively (adopted from Witt et al. 2015). ... 27 Figure 2.15 Vertical profiles of time-averaged air volume fraction for Fr1 = 4.82 at four positions along the length of the hydraulic jump: (a)
x
= 7.14 y1y1; (b)x
= 11.9 y1; (c)x
= 16.67 y1; andx
= 23.8 y1. The open circle symbols are Murzyn et al.’s (2005) measurementsof average void fraction. The dotted and solid curves are Witt et al.’s (2015) 2- and 3-D predictions, respectively (adopted from Witt et al. 2015). ... 27 Figure 3.1 Dimensions of the computational model domain ... 29 Figure 3.2 The finite volume mesh of the computational domain ... 36 Figure 3.3 Initial conditins for computations: (a) the initial condition of the water surface for one of the simulation cases, and (b) the initial condition of the velocity for one of the simulation cases. ... 43 Figure 4.1 Sketch of hydraulic jumps under various inflow conditions (adopted from Chanson 1996, p. 76), where y1 and y2 are the upstream and downstream depths, respectively; is the boundary layer thickness; and v is the flow velocity. ... 45 Figure 4.2 Patterns of two-dimensional air-water flow in hydraulic jump in a rectangular open channel (adopted from Chanson and Brattberg 2000). ... 46 Figure 4.3 Vertical profiles in jump rollers of: (a) air concentration and (b) velocity (Adopted from Chanson and Brattberg 2000). ... 47 Figure 4.4 Time averaged distribution of water volume fraction for simulation case 1, in which the upstream Froude number is Fr1 = 3.1. ... 50 Figure 4.5 Distributions of flow velocity vectors and time-averaged water volume fraction in the hydraulic jump for simulation case 1. The upstream Froude number is Fr1 = 3.1 (Table 4.1 and Table 4.2) ... 51 Figure 4.6 Vertical profile of time-averaged air volume fraction at (x - x1)/y1 = 0.91 for simulation case 1 with Fr1 = 3.1. ... 52 Figure 4.7 Vertical profile of time-averaged air volume fraction at (x - x1)/y1 = 1.7 for simulation case 1 with Fr1 = 3.1. ... 53 Figure 4.8 Vertical profile of time-averaged air volume fraction at (x - x1)/y1 = 3.41 for simulation case 1 with Fr1 = 3.1. ... 54 Figure 4.9 Vertical profile of time-averaged air volume fraction at (x - x1)/y1 = 6.82 for simulation case 1 with Fr1 = 3.1. ... 55 Figure 4.10 Distributions of predicted and measured depth-averaged air volume fraction for simulation case 1. The upstream Froude number is Fr1 = 3.1. ... 56 Figure 4.11 Vertical profiles of the time - averaged longitudinal velocity component below the free surface for simulation case 1, with Fr1 = 3.1. ... 59
Figure 4.12 Time averaged distribution of water volume fraction for simulation case 2, in which the upstream Froude number is Fr1 = 3.8. ... 60 Figure 4.13 Distributions of flow velocity vectors and time-averaged water volume fraction in the hydraulic jump for simulation case 2. The upstream Froude number is Fr1 = 3.8 (Table 4.1 and Table 4.2) ... 61 Figure 4.14 Vertical profile of time-averaged air volume fraction at (x - x1)/y1 = 1.81 for simulation case 2 with Fr1 = 3.8. ... 62 Figure 4.15 Vertical profile of time-averaged air volume fraction at (x - x1)/y1 = 3.81 for simulation case 2 with Fr1 = 3.8. ... 63 Figure 4.16 Vertical profile of time-averaged air volume fraction at (x - x1)/y1 = 7.23 for simulation case 2 with Fr1 = 3.8. ... 64 Figure 4.17 Vertical profile of time-averaged air volume fraction at (x - x1)/y1 = 10.8 for simulation case 2 with Fr1 = 3.8. ... 65 Figure 4.18 Distributions of predicted and measured depth-averaged air volume fraction for simulation case 2. The upstream Froude number is Fr1 = 3.8. ... 67 Figure 4.19 Vertical profiles of the time - averaged longitudinal velocity component below the free surface for simulation case 2, with Fr1 = 3.8. ... 69 Figure 4.20 Time averaged distribution of water volume fraction for simulation case 3, in which the upstream Froude number is Fr1 = 4.4. ... 70 Figure 4.21 Distributions of flow velocity vectors and time-averaged water volume fraction in the hydraulic jump for simulation case 3. The upstream Froude number is Fr1 = 4.4 (Table 4.1 and Table 4.2) ... 71 Figure 4.22 Vertical profile of time-averaged air volume fraction at (x - x1)/y1 = 3.8 for simulation case 3 with Fr1 = 4.4. ... 72 Figure 4.23 Vertical profile of time-averaged air volume fraction at (x - x1)/y1 = 7.59 for simulation case 3 with Fr1 = 4.4. ... 73 Figure 4.24 Vertical profile of time-averaged air volume fraction at (x - x1)/y1 = 11.4 for simulation case 3 with Fr1 = 4.4. ... 74 Figure 4.25 Distributions of predicted and measured depth-averaged air volume fraction for simulation case 3. The upstream Froude number is Fr1 = 4.4 ... 75
Figure 4.26 Vertical profiles of the time - averaged longitudinal velocity component below the free surface for simulation case 1, with Fr1 = 3.1. ... 77 Figure 4.27 Time averaged distribution of water volume fraction for simulation case 4, in which the upstream Froude number is Fr1 = 5.1. ... 78 Figure 4.28 Distributions of flow velocity vectors and time-averaged water volume fraction in the hydraulic jump for simulation case 4. The upstream Froude number is Fr1 = 5.1 (Table 4.1 and Table 4.2) ... 79 Figure 4.29 Vertical profile of time-averaged air volume fraction at (x - x1)/y1 = 3.8 for simulation case 4 with Fr1 = 5.1. ... 80 Figure 4.30 Vertical profile of time-averaged air volume fraction at (x - x1)/y1 = 7.59 for simulation case 4 with Fr1 = 5.1. ... 81 Figure 4.31 Vertical profile of time-averaged air volume fraction at (x - x1)/y1 = 11.4 for simulation case 4 with Fr1 = 5.1. ... 82 Figure 4.32 Vertical profile of time-averaged air volume fraction at (x - x1)/y1 = 15.2 for simulation case 4 with Fr1 = 5.1. ... 83 Figure 4.33 Distributions of predicted and measured depth-averaged air volume fraction for simulation case 4. The upstream Froude number is Fr1 = 5.1. ... 84 Figure 4.34 Vertical profiles of the time - averaged longitudinal velocity component below the free surface for simulation case 4, with Fr1 = 5.1. ... 87 Figure 4.35 the maximum air volume fraction in the shear region in different distances from the jump toe in all 4 test cases in the present study ... 88
List of Tables
Table 2.1 Experimental conditions and measurement locations of Murzyn et al. (2005). ... 16 Table 2.2 Conditions of Chanson’s (2006) experiments. ... 17 Table 2.3 Experimental conditions of weak and steady hydraulic jumps in Lin et al. (2012). 21 Table 4.1 Conditions of CFD simulations, in comparison with the experimental conditions of Chachereau and Chanson (2010). ... 48 Table 4.2 Four selected longitudinal locations for examinations of air volume fraction and vertical velocity distributions. ... 49 Table 4.3 Selected longitudinal locations for the examination of air entrainment and flow velocity pattern sampling lines for simulation case 1 (Tables 4.1 and 4.2). The distance of the jump toe from the inlet is 2.43 m. ... 52 Table 4.4 Selected longitudinal locations for the examination of air entrainment and flow velocity pattern sampling lines for simulation case 2 (Tables 4.1 and 4.2). The distance of the jump toe from the inlet is 2.06 m. ... 62 Table 4.5 Selected longitudinal locations for the examination of air entrainment and flow velocity pattern sampling lines for simulation case 3 (Tables 4.1 and 4.2). The distance of the jump toe from the inlet is 1.16 m. ... 72 Table 4.6 Selected longitudinal locations for the examination of air entrainment and flow velocity pattern sampling lines for simulation case 4 (Tables 4.1 and 4.2). The distance of the jump toe from the inlet is 0.41 m. ... 80
List of Symbols
A flow cross section area (m2)
A cell cross section area related to the direction of v (m2)
C air volume fraction max
C
maximum air volume fraction in the shear region of a hydraulic jump meanC
depth-averaged time-averaged air volume fraction 1 C
k model constant 2 C
k model constant C
k model constant wC
wall function constant tD
turbulent diffusivity of air bubbles in air-water flow (m2/s)E specific energy (m)
e
total energy (J)*
E empirical constant
F bubble frequency defined as the number of detected air bubbles per unit time (Hz)
f force (N)
max
F
maximum bubble frequency (Hz) sF
specific force (m3)g gravitational acceleration (m/s2)
g gravitational acceleration (m/s2)
l
h
energy head loss (m)k kinetic energy of turbulent fluctuations per unit mass (m2/s2)
von Kármán constant
ac
k
turbulent kinetic energy at the wall adjacent cell centroid (m2/s2)eff
k
effective thermal conductivity (W/m.K)L length (m)
m mass flow rate (kg/s)
Nw weber number
p pressure (pa)
0
p
total gage pressure with respect to the operating pressure (pa) sp
hydrostatic pressure with respect to the operating pressure (pa)Q flow discharge (m3/s)
q discharge per unit width (m2/s) Re Reynolds number
ij
S mean strain rate tensor (s-1)
q
s source term of phase q, the mass of phase q which is added to the continuous phase per unit time (kg/s)
t
time (s)U volume flux (m3.s-1.m-2) *
U dimensionless velocity
ac
U
mean velocity of fluid at the wall adjacent cell centroid (m/s)j iu
u temporal average of fluctuating velocities (m2/s2)
V air-water velocity (m/s) or average flow velocity (m/s) v velocity vector (m/s)
v
fluid velocity at the inlet (m/s) maxV
maximum velocity in the x direction (m/s) xV
velocity in the x direction (m/s)Vol fluid volume (m3)
x
horizontal distance from the inlet (m) 1x horizontal distance of the jump toe from the inlet (m)
i
x
position vector in tensor notationy vertical distance measured from the channel bed (m) or depth of water (m) *
y
dimensionless distance from the wall5 . 0
Y
vertical distance from the point whereV
x
V
max/
2
to the channel bed (m) 90y
vertical distance from the point where C0.9 to the channel bed (m) acac
y
the mean ofy
acs of all the cells adjacent to the wall (m) cy
critical depth (m) maxC
Y
vertical distance from the point whereC
C
maxto the channel bed (m)max F
Y
vertical distance from the point whereF
F
maxto the channel bed (m)shear
Y
vertical distance of turbulent shear region upper boundary from the channel bed (m)max V
Y
vertical distance from the point whereV
x
V
max to the channel bed (m)z vertical distance from the free surface level to the centroid of the flow section (m)
volume fraction change in variable
boundary layer thickness (m)
energy dissipation per unit mass (m2/s3) (W/kg)
dynamic viscosity (N.s/m2) kinematic molecular viscosity (m2/s)
T
kinematic eddy viscosity (m2/s)
density of water (kg/m3)
surface tension (kg/m) k
k model constant
k model constant ijw
surface shear stress (N/m2) Subscripts:1 upstream cross section 2 downstream cross section
a related to phase
a
(fluid) in simulationb related to phase b(fluid) in simulation
f related to face of the cell
m model
p related to the pth phase (fluid) in simulation
pq phase p to phase q
pr prototype
q related to the qth phase (fluid) in simulation
qp phase q to phase p
r prototype to model ratio Superscripts:
n previous time step
1
Chapter 1
Introduction
1.1 Background
The hydraulic jump is a flow feature through which the flow of water transfers abruptly from the supercritical to the subcritical condition (Figure 1.1). The sudden change of flow regime is companied by considerable turbulence, loss of flow energy, and entrainment of air mass from above the water surface. Hydraulic jumps are frequently observed to occur in rivers (Figure 1.2), natural streams (Figure 1.3), lakes, coastal water, and man-made water conveyance systems (Figure 1.4). They even occur in kitchen sinks. Hydraulic jumps are classified as rapidly varied flow, as opposed to gradually varied flow. They can be free surface flow or submerged flow below the water surface. Along the length of a hydraulic jump, there is a continuous transition from high flow velocity to low flow velocity, with a corresponding increase in the depth of flow.
The phenomenon of hydraulic jumps is very common in natural water streams and man-made water channels, as shown in Figures 1.1–1.4. Hydraulic jumps have been receiving extensive research attention because of their engineering relevance. However, most of the existing studies of hydraulic jumps have considered the flow as single phase liquid flow, and unrealistically ignored air bubbles [Figure 1.1(b)] as the gas phase.
In Figure 1.2, hydraulic jumps are shown to occur after a dam chute on a natural stream in Australia. It is clear that the river flow was accompanied by river sediment suspension/re-suspension, and channel erosion. The obstacles were built in the flow path to form hydraulic jumps right after the chute. The idea was to make hydraulic jumps in a desired location, and to dissipate the energy of the high speed flow through the jumps. This would prevent the channel downstream of the dam from erosion. In natural streams, hydraulic jumps can occur due to natural obstacles at the bottom of the stream (Figure 1.3), and cause changes to the stream geometry. Such changes may have important implications to the health of the habitats of aquatic species.
The design and operation of water conveyance system must pay close attention to distributions and losses of flow energy. In Figure 1.4, a hydraulic jump is seen immediately downstream of a weir in a water conveyance system. Turbulent motions associated with hydraulic jumps (Figures 1.1 and 1.4) cause considerable loss of flow energy. The losses of flow energy due to the occurrence of hydraulic jump in water conveyance systems affects the efficiency of the systems. Thus, the energy losses need to be taken into account in the design of the systems.
(a)
(b)
Figure 1.1 Photo of the experimental setup in Concordia University’s Water Resources Engineering Laboratory: (a) a rectangular recirculation flume; (b) hydraulic jump as bubbly flow. The direction of flow was from right to left. A control sluice gate allowed a small opening and produced supercritical flow downstream of itself. At the downstream end of the flume, a control tail gate was raised, producing subcritical flow upstream of itself. The two controls created the hydraulic jump. There is an exchange of air mass across the flow surface, with a net entrainment of air mass into the flow, forming air bubbles in the flow.
Figure 1.2 Burdekin dam on the Burdekin River in Queensland, Australia, showing hydraulic jumps induced by obstructions and a gradient change.
(http://www.abc.net.au/news/2016-05-26/the-burdekin-falls-dam-spills-over-as-cyclone/7446962, accessed on February 26, 2017)
Figure1.3 Hydraulic jump on the Naramatagawa River's stream. (https://commons.wikimedia.org/wiki/File:Hydraulic_jump_on_Narama tagawa_River's_stream.JPG, accessed on February 27, 2017)
Figure1.4 Hydraulic jump downstream of a weir in an open channel. (http://www.itrc.org/projects/flowmeas.htm, access on september 2, 2016)
The transition of shallow, fast flow to deep, slow flow through a hydraulic jump features extreme chaos, large scale turbulence, surface waves, and spray. Because of the highly turbulent flow condition in hydraulic jumps, air mass immediately above the flow surface will enter the flow. Thus, the flow becomes two phase flow, with water as the liquid phase and air as the gas phase. The exchange of air mass between the atmosphere and water is known as air entrainment (or bubble entrainment or aeration). The presence of air in the flow makes it difficult to determine the exact location of the interface between the flowing fluids and the atmosphere above. Besides, there is a continuous exchange flux of liquid and air between the flow and the atmosphere. The resulting air-water mixture consists of both air pockets within water and water droplets surrounded by air. There are also spray, foam, and complex air water structures. In all situations, the flow is constituted of both air and water.
On the basis of the Reynolds number (Reynolds 1883), fluid flows are distinguished into laminar (or smooth) flow and turbulent (or chaotic) flow. There are significant challenges in
realistically predicting turbulent flows because of their irregular chaotic motions, strong mixing properties, and a broad spectrum of length scales. Knowledge of hydraulic jumps with the above mentioned complexities is far from complete.
In a hydraulic jump, at any point, the fluid velocity changes continuously in both magnitude and direction (Chanson 1996, p. 4). The air–water flow of a hydraulic jump includes three distinct regions:
a) A recirculation layer at the top where recirculating flow and large eddies occur in this layer. This layer is characterised by the development of large-scale vortexes and bubble coalescence with a foam layer at the free surface with large air polyhedral structures; b) A turbulent shear layer or air diffusion layer with air bubbles of smaller sizes and high air
content;
c) An impingement jet region at the bottom of shear layer which has a velocity distribution similar to the upstream flow and less or no air bubble can be seen in this region.
The extremely turbulent flow is associated with large scale vortexes. These vortexes and the non-stable dynamic velocity produce significant pressure pulsation and develop a wavy flow. All these characteristics results in a wild and erosive flow, which can cause erosion on non-protected open channel beds and walls, and transport and can mix a large amount of sediments with the flowing water. The resultant high turbidity will deteriorate water quality. Also, the turbulent and wavy flow produces a clearly detectable sound and converts a relatively large amount of energy into heat.
In water conveyance settings, the study of hydraulic jumps in open channels has many important applications. Examples include:
a) Reduction of excessive energy of flowing water for the safety of hydraulic structures. In certain part of an open channel system with a steep bed slope, the flow gains velocity. At high velocity, the flow becomes erosive or destructive. The use of hydraulic jump in such situation will dissipate the extra amount of flow energy and hence avoid destructive effects; b) Efficient mixing of chemicals in water treatment plants. The use of hydraulic jump represents a very suitable method for mixing chemical substances required in the process of water treatment. The macro-scale vortexes in the hydraulic jump increase the efficiency of mixing procedures;
c) Discharge measurements. In Parshall flumes (Chow 1959), a hydraulic jump is produced for determining flow discharge by measuring the depth of water at certain locations. Clearly, there are beneficial applications of hydraulic jump as well as undesirable consequences of their occurrences such as erosion in erodible channels or turbulent disturbances in water conveyance systems. There is a need for improved understanding of various aspects of the hydraulic jump phenomenon.
Classic studies solved the problem of hydraulic jumps as one-dimensional flow using the continuity and momentum principles. The solution approach is overly simplified, without dealing with turbulence. The present knowledge of the turbulent flow field is fairly limited, especially under environmental and geophysical flow conditions (Chanson 2009). After extensive research, the hydraulic jump phenomenon remains a fascinating flow motion, and the present knowledge is insufficient in several aspects, including air entrainment, turbulence, and undular flow.
This present research work focuses on distributions of air entrainment in hydraulic jumps at high Reynolds numbers. The consideration of high Reynolds numbers will better reflect real-world conditions under which hydraulic jumps occur in open channels and water conveyance systems. This represents an extension of the previous studies, most of which used laboratory or numerical models of small length scales. The results from these studies have inevitably suffered from a scale problem. In the laboratory experiments reported in the literature, the Reynolds number was typically much smaller than real-world jumps in irrigation and water conveyance systems. The use of high Reynolds numbers in this study will improve the relevance to reality.
1.2 Specific aims of this research work
The specific objectives of this research work are as follows:
▪ To produce the hydraulic jump as two-phase flow in a large-scale channel at high Reynolds numbers. This will minimise scale effects.
▪ To quantify the distributions of air volume fraction in hydraulic jump rollers in a range of Froude numbers. Air volume distributions in the flow give rise to non-hydrostatic pressure. A good understanding of the distributions represents a significant improvement from the traditional simplification of hydrostatic pressure (no air bubbles) field in hydraulic jumps.
▪ To establish suitable computational procedures for predicting air entrainment in hydraulic jumps. The suitability is to be confirmed by comparing air entrainment between the computer model and existing laboratory experiments.
▪ To quantify the average percentage of air in hydraulic jumps in a range of Froude numbers.
▪ To reveal the vertical structures of flow velocity along hydraulic jump rollers in a range of Froude numbers.
1.3 Scope of this research work
To achieve the above-mentioned objectives, the rest of this thesis is organized as follows: Chapter Two will introduce the classical theory and basic characteristics of hydraulic jumps. The chapter will discuss the similitude of models for hydraulic jumps, and will provide highlights of the previous experimental and numerical studies of air entrainment in hydraulic jumps. Outstanding issues will be outlined.
Chapter Three will discuss on modelling methodologies. The chapter will give a description of the model domain and geometric setup. Discussions of the modelling methodologies will cover the volume of fluid method, and the standard 𝑘-ε model for turbulence closure. The chapter will provide details of computational mesh configuration for hydraulic jump simulations, treatment of open boundary conditions, and specification of initial conditions.
Chapter Four will present the computational results, along with a comparison with available experimental data. The results will the flow velocity field, distributions of air volume fraction, flow surface profile, and turbulence kinetic energy. The characteristics of the above-mentioned quantities at different Froude numbers will be discussed in details. The comparison will make use of the experimental data from Chachereau & Chanson (2010). Through the comparison, we will investigate the effects of Reynolds number on air entrainment.
In Chapter Five, conclusions will be drawn, and suggestions for future studies of hydraulic jumps as two-phase flow will be outlined.
1.4 Highlights of research contributions
New contributions from this research work are highlighted below:
(1) Reliable predictions of hydraulic jumps as two-phase bubbly flow at high Reynolds numbers, which has not been achieved in previous studies.
(2) Quantitative details of the flow field and air entrainment distributions that are difficult to measure from laboratory experiments.
(3) An improved understanding of the effects of the Reynolds number on air entrainment in oscillating and steady hydraulic jumps. This is of practical importance to the design of appropriate laboratory setup for hydraulic jump experiments.
(4) The establishment of suitable computational procedures for numerical simulations of hydraulic jumps and the determination of time averaged variables related to the jumps.
Chapter 2
Review of the Pertinent Literature
2.1 The classical theory of hydraulic jumps
Open-channel flow can change from a subcritical state to a supercritical state and vice versa, in response to certain changes in channel geometry or flow boundary conditions or both. Changes from a subcritical to supercritical state usually occur rather smoothly via critical depth. However, changes from a supercritical to subcritical state occurs abruptly through a hydraulic jump (Figure 2.1). The depth of flow changes from y1 to y2, known as the initial depth of flow before the jump and the sequent depth of flow after the jump.
Figure 2.1 Specific energy diagram, hydraulic jump and specific force diagram (adopted from Houghtalen et al. 2017).
Hydraulic jumps are highly turbulent, with complex internal flow patterns, and dissipate considerable flow energy. Thus, we expect the flow energy to be much lower downstream of a jump than upstream, but have no prior knowledge about the actual amount of energy losses in the jump. Thus, it is difficult to solve the problem of hydraulic jumps by directly using the energy principle. The classical theory of hydraulic jump uses the momentum principle, expressed as (Henderson 1966, p. 74): 2 2 2 1 1 1 A z gA Q A z gA Q (2.1) where the subscripts 1 and 2 refer to the flow sections before and after the jump, respectively; Q
to the centroid of the flow section; and A is the flow area. Gravity is an important parameter for the problem of hydraulic jumps because the flow has a free surface (Figure 2.1).
Equation (2.1) has assumed: a) The frictional forces at the channel bed and on the channel sidewalls are negligible; b) there are no external forces other than pressure forces; c) the channel has a horizontal bed; and d) the flow is incompressible. The sum of the two terms on each side of Equation (2.1) is known as the specific momentum
F
s. Thus, the two flow sections before and after the hydraulic jump, respectively, have the same specific momentum:2
1 s
s
F
F
(2.2) For given hydraulic conditions and channel geometry, Equation (2.2) can be solved analytically to yield the relationship between y1 and y2 (Figure 2.1). Then, the energy equation can be used to determine the amount of flow energy losses in the hydraulic jump.2.2 Hydraulic jump in rectangular channels
For hydraulic jumps in a horizontal, rectangular channel, Equation (2.1) is reduced to: ) 2 ( ) 2 ( 2 2 2 2 2 1 1 2 y gy q y gy q (2.3) where q is the discharge per unit width of channel. Solving Equation (2.3) yields (Henderson 1966, p. 69):
)
1
8
1
(
2
2 1 1 2
Fr
y
y
(2.4) or(
1
8
1
)
2
2 2 2 1
Fr
y
y
(2.5) where Fr1 is the upstream Froude number; and Fr2 is the Froude number after the jump. Fr1 is given by: 1 1 1 gy V Fr (2.6)where V1 is the depth of flow at upstream, which is related to q and y1 as V1 = q/y1. Similarly, Fr2 is given by: 3 2 2 gy q Fr (2.7)
The Froude number is an important parameter for the hydraulic jump phenomenon, and therefor the gravity plays a significant role in studying the hydraulic jump phenomenon.
This well-known dimensionless parameter in free surface flow represents the ratio of the inertial force in the flow to gravity force.
For a given initial depth, y1, of flow before the jump (Figure 2.1), the sequent depth, y2, of flow after the jump can be determined from Equation (2.4), and vice versa from Equation (2.5). The amount of energy head losses, hl, in the hydraulic jump is given by:
) 2 ( ) 2 ( 2 2 2 2 1 2 1 2 y gy q y gy q hl (2.8) or equivalently 2 1 3 1 2 4 ) ( y y y y hl (2.9)
2.3 Basic characteristics of hydraulic jumps in rectangular channels
Akan (2011, p. 238) provided a description of a series of tests by U.S. Bureau of Reclamation (1987) to study the hydraulic jumps. The results of these tests show that the shape, form and characteristics of hydraulic jumps depend on Fr1 (Equation 2.6). In Figures 2.2(a)-2.2(d), hydraulic jumps are classified based on Fr1.
(a) Form A: 1.7 < Fr1 < 2.5
(b) Form B: 2.5 < Fr1 < 4.5
(d) Form D: Fr1 > 9.0
Figure 2.2 Classifications of hydraulic jumps according to the upstream Froude number Fr1 (adopt from US Bureau of Reclamation 1987).
For 1 < Fr1 < 1.7, the approach flow depth is only slightly less than the critical depth. From this transition of a supercritical to subcritical stage, the flow changes gradually with a very slight turbulent water surface. Some small rollers begin to form on the surface as Fr1 approaches 1.7. The change becomes more intense with increasing upstream Froude number. Except the existence of surface rollers, relatively smooth flows predominate at Froude numbers up to about 2.5. Hydraulic jumps with Froude numbers between 1.7 and 2.5 are characterize as form A [Figure 2.2(a)].
For 2.5 < Fr1 < 4.5, the hydraulic jump has the characteristics of oscillating flow. This oscillating flow has undesirable and sometimes significant surface waves that carry far downstream. The hydraulic jumps for this range of upstream Froude numbers are classified as form B [Figure 2.2(b)].
For 4.5 < Fr1 < 9.0, the hydraulic jump is well-balanced and stable in place. Turbulence is limited to the main body of the hydraulic jump, and the water surface downstream is relatively smooth. The hydraulic jumps in this range of upstream Froude numbers are considered as form C and are called steady jumps [Figure 2.2(c)].
For Fr1 > 9.0, the turbulence through the hydraulic jump and the surface rollers become significantly active. This causes a jump with a rough water surface and with strong water waves carry downstream from the jump. This kind of hydraulic jump is called strong jump and is classified as form D [Figure 2.2(d)].
2.4 Similitude of models for hydraulic jumps
General discussions about hydraulic similitude can be found in Houghtalen et al. (2017). Hydraulic modelling of a prototype must ensure three types of similarities as listed below: geometric similarity, kinetic similarity, and dynamic similarity. The geometric similarity is the similarity of form. It means that when a prototype size is reduced, the homologous lengths must have a fixed ratio between the hydraulic model and prototype. In this connection, there are three physical factors involved in geometric similarity: length, area, and volume.
The kinematic similarity is the similarity of motion. This similarity will be achieved when the homologous moving particles moving through geometrically similar paths have similar velocity ratios between the hydraulic model and prototype. This type of similarity involves two factors: length and time.
The dynamic similarity is the similarity of forces active in the motion. This similarity will be achieved when homologous forces involved in the motion have a fixed ratio between the hydraulic model and prototype, or:
r m pr f f f (2.10) The study of air entrainment in the hydraulic jump phenomenon involves several kinds of forces in action. The dynamic similarity in hydraulic jump studies requires that the ratio of these forces be kept the same between the model and the prototype. These force ratios are discussed in following sections.
2.4.1 Viscous force – Reynolds number law
Inertial forces are known to always affect water in motion such as in hydraulic jumps (Figure 2.1). Consider that the inertial forces and viscous forces as two kinds of forces act on moving homologous particles in the hydraulic model of hydraulic jumps and prototype hydraulic jumps. The ratio of the two types of forces is defined by Reynolds number law. For applications to hydraulic jumps (Figure 2.1), the Reynolds number base on variables of the approach flow can be expressed as:
1 1
1
where is the kinematic viscosity of water. If the inertial force and the viscous force are the main forces governing the fluid motion, one must keep the same Reynolds number between the model and the prototype (Houghtalen et al. 2017, p. 379).
2.4.2 Gravity force – Froude number law
Gravity is an important parameter for the free surface phenomenon of hydraulic jumps. The ratio of the inertial force to gravity force is defined by the Froude number. The Froude number based on the upstream flow is given in Equation (2.6). To ensure that the behaviour of the model hydraulic jumps reflects that of the prototype jumps, one must retain the same value for the Froude number between a hydraulic model of hydraulic jumps and prototype of hydraulic jumps.
2.4.3 Surface tension – Weber number law
In Houghtalen et al. (2017, p. 383), surface tension is described as a measure of energy level on the surface of a liquid body. Hydraulic jumps are turbulent flow, which entrains air from above the free surface. The entrainment process results in continuous mixing of air and water. This gives rise to surface tension force in hydraulic jumps.
The Weber number defines the ratio of inertial force to surface tension force:
VL
Nw (2.12) where ρ is the density of the fluid, V is a velocity scale, L is a length scale, and σ is surface tension per unit length. One ought to keep the same value for the Weber number between the hydraulic model and the prototype.
In this research work, the theory for hydraulic jump predictions, to be presented in the next chapter, considers inertial forces, viscous forces, gravity force, and surface tension. All these forces are involved in the governing equations of fluid motions. Thus, to ensure the relevance of predicted hydraulic jump behaviour to a real-world hydraulic jump, it is desirable to use realistic values for the Reynolds number, the Froude number, and the Weber number in the computations. 2.5 Experimental studies of air entrainment in hydraulic jump
Rajaratnam (1962) is probably the first researcher making laboratory measurements of air volume fraction in hydraulic jumps. Resch and Leutheusser (1972) obtained measurements of air entrainment and air volume fraction in the bubbly flow region of a hydraulic jump, using a
hot-film probe. For the first time, they reported the effects of upstream flow conditions on hydraulic jumps. They suggested that the air entrainment process, momentum transfer and energy dissipation are strongly affected by the inflow conditions. In an experimental study, Chanson and Qiao (1994) focused on air-water properties in partially developed hydraulic jumps. Chanson (1995b) reported new experimental data of air bubble diffusion in turbulent shear flows. The author investigated two flow scenarios: a vertical supported jet, and a horizontal hydraulic jump. As an extension of the previous experimental studies, Chanson and Brattberg (2000) investigated air-water flow properties in the shear regions (Figure 2.4) of hydraulic jumps, under the conditions that the upstream Froude number had values of Fr1 = 6.33 and 8.48. Almost all of the above-mentioned experimental studies used a horizontal, rectangular channel. A definition diagram of hydraulic jump in such a channel is shown in Figure 2.3.
Figure 2.3 Definition diagram of hydraulic jump experiments in a rectangular channel (adopted from Chanson and Brattberg, 2000). Note that V represents the flow velocity in the x -direction.
Using a dual-tip optical fibre probe, Murzyn et al. (2005) obtained measurements of air volume fractions, bubble frequencies and bubble sizes in hydraulic jumps for four different cases of the upstream Froude number Fr1. The measurements were from a large number of points throughout the jump. The experimental conditions of Murzyn et al. (2005) are summarised in
Table 2.1. Note that y1 and y2 are the upstream and downstream depths of flow, respectively (Figures 2.1-2.3); V1 is the upstream flow velocity (Figure 2.1); x1 is the horizontal distance between the sluice and the toe of the hydraulic jump in question, and x is the horizontal distance measured from the sluice gate (Figure 2.3). Murzyn et al. (2005) selected four to five measurement locations along the length of the hydraulic jump, given in terms of (x–x1)/y1.
Table 2.1 Experimental conditions and measurement locations of Murzyn et al. (2005). Experiment y1 y2 V1 Fr1 Re1 x1 (x - x1)/y1 (m) (m) (m/s) - - (m) - 1 0.059 0.138 1.50 2.0 88500 0.36 0.85, 1.70, 2.54, 4.24 2 0.046 0.137 1.64 2.4 75440 0.28 2.17, 4.35, 6.52, 8.70, 10.90 3 0.032 0.150 2.05 3.7 65600 0.34 4.69, 7.81, 15.60, 20.30, 25.00 4 0.021 0.133 2.19 4.8 45990 0.36 7.14, 11.90, 23.80, 31.00, 38.10 Several studies (Resch and Leutheusser 1972; Chanson 1995a,b) reported that distributions of air concentration C (or equivalently air volume fraction) exhibit a peak in the turbulent shear region, as illustrated in Figure 2.4. Chanson (1995a, 1996) related C to a solution of the diffusion equation, which is given by:
1 1 2 1 1 1 1 max / 4 / / exp max y x x D y Y y y y V C C t C for y < Yshear (2.13) where Cmax is the maximum air content in the turbulent shear layer region measured at a distanceYCmax from the bottom; V1 is the free-stream velocity of the inflow; Dt is a turbulent diffusivity;
Figure 2.4 Vertical distributions of air concentration (air volume fraction) and bubble frequency in hydraulic jump rollers (from Chanson and Brattberg 2000).
Chanson (2006) experimentally investigated the entrainment of air bubbles in the developing region of a hydraulic jump under partially developed inflow conditions. The experiments were conducted in two flumes of similar geometry but different widths: One was narrow, and the other was wide. The idea was to assess the effects of channel width (scale) on air entrainment. The experimental conditions are given in Table 2.2. An example of results is shown in Figure 2.5, where F is air-bubble count rate (Hz) or bubble frequency (defined as the number of detected air bubbles per unit time). There is a good agreement between the theory (Equation 2.13; Figure 2.5, the solid curve) and experimental data (Figure 2.5, the solid squares) of Chanson (2006).
Table 2.2 Conditions of Chanson’s (2006) experiments.
Flume width B y1 V1 Fr1 Re1 x1 Remark (m) (m) (m/s) - - (m) 0.25 0.0133 1.86 5.1 24738 0.5 Narrow channel 0.25 0.0129 3.00 8.4 38700 0.5 0.25 0.0290 2.67 5.0 77430 1.0 0.25 0.0245 3.90 7.9 95550 1.0 0.50 0.0265 2.60 5.1 68900 1.0 Wide channel 0.50 0.0238 4.14 8.6 98532 1.0
Figure 2.5 Distributions of air volume fraction and bubble count rate, measured at 0.3 m downstream of the jump toe by Chanson (2006). The upstream Froude number was Fr1 = 8.6. The channel was 0.5 m wide (adopted from Chanson 2006).
Using Chanson’s (2006) narrow channel (Table 2.2), Gualtieri and Chanson (2007) carried out a further experimental study of the vertical distribution of air volume fraction and bubble count rate in hydraulic jumps. The experiments produced results for the upstream Froude number in the range of Fr1 = 5.2 to 14.3. In Figure 2.6, as an example, the vertical distribution of air volume fraction at (x - x1)/y1 = 11.63 is plotted. A comparison of the result with those reported in Chanson (1995 a) and in Chanson and Brattherg (2000) leads to the following observations:
▪ There is a decrease in the maximum air content in the turbulent shear layer with increasing distance from the jump toe. The data points appear to follow closely both power law and exponential decay functions, as suggested by Chanson and Brattberg (2000) and Murzyn et al. (2005).
▪ There is an exponential decay in the maximum bubble frequency (Figure 2.4) with increasing distance from the impingement point.
▪ The decay of the maximum air content with increasing distance from the impingement point is lower at higher Fr1.
▪ The decay of the maximum number of bubble impacting the probe is lower at higher Fr1.
Figure 2.6 Vertical profile of air volume fraction at (x - x1)/y1 = 11.63. The experimental conditions were Fr1 = 8.37, Re1 = 38410, and y1 = 0.0129 m (adopted from Gualtieri and Chanson 2007).
The measurements of air volume fraction in Chanson (2006) and in Chanson and Murzyn (2008) correspond to similar experimental conditions. The upstream Froude numbers were identical. The Reynolds numbers were between Re1 = 24738 and 98532 (Table 2.2). A comparison of measured air-volume-fraction and bubble-count-rate distributions between the two studies reveals drastic scale effects in relatively small hydraulic jumps. In their comparative analysis, Chanson and Murzyn (2008) demonstrated quantitatively that a dynamic similarity of two-phase flows in hydraulic jumps could not be achieved with a Froude similitude. At Reynolds numbers below 105, the experimental data show some viscous scale effects on the rate of air entrainment and air-water interfacial area. In Figure 2.7, a comparison of air volume fraction between Chanson (2006) and Chanson and Murzyn (2008) is shown for Fr1 = 5.1 and (x - x1)/y1= 8.
Figure 2.7 Comparison of air volume fraction in hydraulic jumps between Chanson (2006) (with Re1 = 24738 and 68900) and Chanson and Murzyn (2008) (with Re1 = 38576). The upstream Froude number is Fr1 = 5.1. The distance is (x - x1)/y1 = 8 (adopted from Chanson and Murzyn 2008).
Using Chanson’s (2006) wide channel (Table 2.2), Chachereau and Chanson (2010) performed experiments of hydraulic jumps with the inflow Froude number in the range of Fr1 = 2.4 to 5.1. They investigated fluctuations in the free surface and turbulence, and air-water flow properties. They concluded that vertical profiles of air volume fraction had two characteristic regions: a shear layer region in the lower part of the flow, and an upper free-surface region above.
Chachereau and Chanson (2010) compared their results for Fr1 = 5.1 with Chanson’s (2006) results for the same value of Fr1 but smaller values of Re1. In this comparative analysis, the Reynolds number ranges from Re1 = 24738 (Table 2.2) to 125400. It was shown that the Froude similitude was not satisfied in a hydraulic jump for Fr1 = 5.1 within the range of Reynolds numbers. The data of air volume fraction obtained with Reynolds numbers below 40000 could not be scaled up to larger Reynolds numbers. The bubble count rate, turbulence properties, and bubble chord exhibited monotonic trends with increasing Reynolds numbers. The implication was that the results could not be extrapolated to large-size prototype structures without significant scale effects.
Using particle image velocimetry (PIV) and bubble image velocimetry (BIV) techniques, Lin et al.(2012) measured the flow structures and turbulence statistics of three steady hydraulic jumps. The upstream Froude number was in the range of Fr1 = 4.51–5.35. Measurements were made from both the non-aerated and aerated regions of the hydraulic jumps, and were validated using laser Doppler velocimetry (LDV) and tracking bubble trajectories. Lin et al. (2012) also obtained measurements from a weak jump, with a Froude number of Fr1 = 2.43, aiming to examine the differences between weak and steady hydraulic jumps. The experimental conditions of Lin et al. (2012) are listed in Table 2.3.
Table 2.3 Experimental conditions of weak and steady hydraulic jumps in Lin et al. (2012).
Experiment y1 y2 V1 Fr1 Re1 Remark (m) (m) (m/s) - - 1 0.0192 0.0570 1.063 2.43 20410 Weak jump 2 0.0195 0.0115 1.973 4.51 38474 Steady jump 3 0.0200 0.0132 2.216 5.00 44320 4 0.0195 0.0138 2.337 5.35 45572
To the best of our knowledge, Wang and Chanson (2015) is the most recent experimental study of air entrainment in a hydraulic jump. Their experiments covered a wide range of Froude numbers (3.8 < Fr1 <10.0) and Reynolds numbers (35800 < Re1 < 164000). The authors investigated non-intrusively fluctuations in the free surface and roller position, using a series of acoustic displacement meters. They reported the characteristic frequencies of the fluctuating motions, some major roller surface deformation patterns, air-water flow properties, air volume fraction and bubble count rate in the rollers, and interfacial velocity distributions. Wang and Chanson (2015) used Chanson’s (2006) wide channel (Table 2.2). An example of vertical distributions of air volume fraction is shown in Figure 2.8.
Figure 2.8 Vertical distributions of air volume fraction at a series of positions along the length of the hydraulic jump. The flow conditions were: Q = 0.0347 m3/s, y
1 = 0.0206 m, x1 = 0.8 3m, Fr1 = 7.5, and Re1 = 68000 (adopted from Wang and Chanson 2015).
2.6 Numerical studies of air entrainment in hydraulic jumps
A review of the literature shows a very limited number of numerical studies of hydraulic jumps as two-phase open-channel flow. This section discusses the previous numerical studies dealing with air entrainment and air volume fraction distributions in hydraulic jumps.
Ma et al. (2011) simulated hydraulic jumps using a subgrid air entrainment model in conjunction with 3D two-fluid models of bubbly flow (Reynolds-averaged equation model and Detached Eddy Simulation model). The upstream Froude number was Fr1 = 1.98 and the Reynolds number was Re1 = 88500. They predicted air volume fraction distributions at a number of locations downstream of the jump toe and compared the results with the measurements of Murzyn et al. (2005). In Figure 2.9, simulated and measured air volume fraction profiles are plotted as a function of the normalised vertical coordinate, where y/y1=0 denotes the bottom of the channel.
Figure 2.9 Distributions of air volume fraction predicted with the Reynolds-averaged equation model (left) and Detached Eddy Simulation model (right) at Fr1 = 1.98 at (xx1)/y1= 0.85, 1.7 and 2.54 (from top to bottom), in comparison with the measurements of Murzyn et al. (2005). The middle column presents DES results accounting for contributions from bubbles, while excluding those from the wavy interface (adopted from Ma et al. 2011).
Regarding the Reynolds-averaged equation model, Ma et al. (2011) noted that the lower half of the air volume fraction profiles, corresponding to the shear layer region, matched the experimental data quite well, but the upper half of the profiles, correspond to the roller region (Figure 2.1), did not. They suggested that the Detached Eddy Simulation results matched the measurements well both in the lower shear layer and in the upper roller region (Figure 2.9).
Recently, Xiang et al.(2014) presented an Eulerian multi-fluid model for investigating flow structures of hydraulic jumps. They obtained explicit solutions to the phasic distribution of fluids through interfacial momentum transfer models. Air ingestion at the jump toe was handled by a sub-grid air entrainment model. The location of the free surface was captured using a compressive
VOF model. Xiang et al. (2014) considered mechanistic coalescence and breakage kernels in the calculations. Their idea was to better represent the evolution of air bubble size in the subcritical flow region. The control parameters of their model simulations are listed in Table 2.4. Xiang et al. (2014) compared their numerical results to experimental data in Chachereau and Chanson (2010) and Lin et al. (2012).
Table 2.4 Physical parameters of three selected flow cases in Xiang et al. (2014).
Case Fr1 y1 y2 Q Re1 Remark
- (m) (m) (m3/s) -
1 3.1 0.0440 0.174 0.0446 89000 Chachereau and Chanson’s
(2010) experiment 2 5.1 0.0395 0.254 0.0627 130000
3 4.5 0.0195 0.115 0.0192 38400 Lin et al.’s (2012) experiment
Xiang et al. (2014) assumed two-dimensional steady state flows in all simulations. Their computational domain consisted of 28,000 non-uniform cells (Figure 2.10). The simulations used ANSYS CFX12. Air entrainment and complete merging model were implemented using CFX Expression Language (CEL). The associated source term for the multiple-size-group model was incorporated into the simulations. CEL allows users to define inputs as variables, capture outputs as variables, and perform operations on those variables.
Figure 2.10 Model domain, mesh and boundary conditions in Xiang et al. (2014).
Xiang et al. (2014) predicted distributions of water superficial velocity vectors for Case 3 (Figure 2.11), air volume fractions in hydraulic jump rollers for Cases 1 and 2 (Figure 2.12), and air volume fractions at 3 different locations downstream of the jump toe for Case 1 (Figure 2.13).
Figure 2.11 Predicted water velocity vectors for Case 3 (Table 2.4) in Xiang et al. (2014).
Figure 2.12 Contours of water and air volume fraction for: (a) Case 1, and (b) Case 2 in Xiang et al. (2014).
Figure 2.13 Distributions of air volume fraction for Case 1 at axial sections (xx1)/y1= 0.91 m (panel a), 1.7 (panel b), and 3.41 (panel c). Note that y is the vertical position (adopted from Xiang et al. 2014).
Witt et al. (2015) numerically simulated air–water flow characteristics in hydraulic jumps in an open channel, using setup corresponding to the laboratory experiments of Murzyn et al. (2005), which was discussed in the previous section. Witt et al.’s (2015) simulations used OpenFOAM
(Jasak 2009), and produced unsteady flow field in two and three dimensions. They solved the evolving free surface using InterFoam (a VOF solver), and located the free surface using air volume fraction of 0.5 as a threshold. Witt et al. (2015) obtained time average results. In this connection, a comparison of relative errors between sampling times of 1, 5, 10, 15 and 20 seconds shows that a sampling time of 15 seconds gave the lowest relative error. Witt et al. (2015) reported distributions of time averaged volume fraction (Figure 2.14) and vertical profiles of average air volume fraction (Figure 2.15), with a comparison to the experimental data of Murzyn et al. (2005).
Figure 2.14 Distributions of predicted volume fraction for a 2-D simulation with the upstream Froude number Fr1 = 4.82. Panel (a) shows an instantaneous distribution. Panels (b), (c), (d), (e), and (f) show the time-averaged distributions over the durations of 1, 5, 10, 15, and 20 s, respectively (adopted from Witt et al. 2015).
(a) (b) (c)
(d)
Figure 2.15 Vertical profiles of time-averaged air volume fraction for Fr1 = 4.82 at four positions along the length of the hydraulic jump: (a)
x
= 7.14 y1y1; (b)x
= 11.9 y1; (c)x
= 16.67 y1; andx
= 23.8 y1. The open circle symbols are Murzyn et al.’s (2005) measurements of average void fraction. The dotted and solid curves are Witt et al.’s (2015) 2- and 3-D predictions, respectively (adopted from Witt et al. 2015).2.7 Summary
Ideally, laboratory and computer modelling of hydraulic jumps should keep the same values of the Reynolds number, the Froude number, and Weber number as the prototype. In most of cases, laboratory modelling studies satisfied geometrical similarity, but not simultaneously Froude number, Reynolds number and Weber number similarities. With the same fluids (air and water) in model and prototype, the process of air entrainment is adversely affected by significant scale effects in small size models (Chanson 2006).
The use of small length scales is a common limitation of previous experimental and numerical studies of air entrainment in hydraulic jumps. There is a need for further studies using relatively large and more practical dimensions so as to satisfy the Froude number, Reynolds number and Weber number similarities at the same time.
Previous studies have been limited to hydraulic jumps of small dimensions. The behaviour may not reflect truly the behaviour of hydraulic jumps in real-world open channels and water conveyance systems. The largest upstream Reynolds number (Equation 2.11) was perhaps 125400 reported in Xiang et al. (2014) who simulated hydraulic jumps and 164000 reported in Wang and Chanson (2015) who made laboratory measurements. Xiang et al.’s (2014) simulations were limited to steady state flow conditions, as opposed to transient conditions.
In summary, the previous studies of hydraulic jumps have rarely reached flow conditions with upstream Reynolds numbers exceeding 106. A knowledge gap exists with regard to distributions of air entrainment in hydraulic jumps of large and practical dimensions. This corresponds to large Reynolds numbers. The need for simulations of hydraulic jumps under conditions of transient motions and large Reynolds numbers have motivated this research work.
Chapter 3
Modelling Methodology
The aim of this study is to simulate four different hydraulic jumps in an open channel with four different upstream Froude numbers under transient condition and to predict air entrainment in these hydraulic jumps. The results will be compared with those of Chachereau and Chanson (2010) who used a physical model to study air entrainment in four different hydraulic jumps at (smaller dimensions and smaller Reynolds numbers). These computational simulations use ANSYS 17.1 (Fluent). In this chapter, a description of the simulation domain will be presented. The governing model equations used for the simulations, boundary conditions and initial conditions will be presented.
3.1 Model domain and geometry
In this section, the geometry and general shape of the model domain used in all the simulations in this study will be presented. The model domain consists of a channel with a length of 6 m and a height of 2.2 m. The channel is horizontal and has two inlets and one outlet: one water inlet, one air inlet at the top, and one outlet to allow water and air to leave the domain during the simulations. All the simulations are two dimensional (Figure 3.1).
3.2 Volume of fluid (VOF) model theory
The VOF model (Hirt and Nichols 1981) can model two or more fluids which are not mixable by solving momentum equation, energy equation and tracking the volume fraction of each fluid throughout the domain. Modelling of open channel flow is a typical application of this method. 3.2.1 Steady – state and transient VOF calculations
The model in ANSYS Fluent is generally used to compute a time-dependent or transient solution, but in some cases that involve steady state flow, it is possible to perform steady-state calculations.
The VOF model relies on the fact that two or more fluids (or phases) are not interpenetrating. For each phase in the calculations, a variable is introduced to represent the volume fraction of the phase in each computational cell. The sum of all volume fractions of all phases is one in each computational cell. The fields for all variables and properties are shared by the phases. In other words, all the fluid properties are represented in form of volume-averaged values, as long as the volume fraction of each of the phases is known at each location. Thus, the variables and properties in any cell in computational domain are either completely representative of one of the phases, or representative of a mixture of the two phases (air and water in this study) or more, depending upon the volume fraction values. In other words, if the qthfluid’s volume fraction in the cell is denoted asαq, then the following three conditions are possible:
(a) αq = 0: The cell is empty of the qth fluid (b) αq =1: The cell is full of the qth fluid
(c) 0 < αq < 1: a certain fraction of cell (αq) is filled with qth fluid.
Based on the local value of αq, the appropriate properties and variables will be assigned to each control volume within the domain (Hirt and Nichols 1981) (Fluent 2013).
3.2.2 Volume fraction equation
The tracking of the interface(s) between the phases is accomplished by solving the continuity equation for the volume fraction of one (or more) of the phases. For the qth phase (air in this study), this equation has the following form (Walters and Wolgemuth 2009) (Fluent 2013):
n